add val gen

scripts/distill_hunyuan.md

@@ -0,0 +1,5 @@
+## How to Distill Hunyuan 
+
+python scripts/download_hf.py --repo_id FastVideo/hunyuan --local_dir data/hunyuan --repo_type model
+python scripts/download_hf.py --repo_id FastVideo/Hunyuan-Distill-Data --local_dir data/Hunyuan-Distill-Data --repo_type=dataset
+

scripts/distill_hunyuan.sh

@@ -23,7 +23,7 @@ torchrun --nnodes 1 --nproc_per_node 8\
     --learning_rate=1e-6\
     --mixed_precision="bf16"\
     --checkpointing_steps=64\
-    --validation_steps 32\
+    --validation_steps 64\
     --validation_sampling_steps "2,4,8" \
     --checkpoints_total_limit 3\
     --allow_tf32\

scripts/preprocess_hunyuan_data.sh

@@ -4,9 +4,9 @@ MODEL_PATH="data/hunyuan"
 MODEL_TYPE="hunyuan"
 DATA_MERGE_PATH="data/Mixkit-All-Clips/merge.txt"
 OUTPUT_DIR="data/Hunyuan-Mixkit-Data"
-
+VALIDATION_PATH="assets/prompt.txt"
 # torchrun --nproc_per_node=$GPU_NUM \
-#     ./fastvideo/data_preprocess/finetune_data_VAE.py \
+#     fastvideo/data_preprocess/preprocess_vae_latents.py \
 #     --model_path $MODEL_PATH \
 #     --data_merge_path $DATA_MERGE_PATH \
 #     --train_batch_size=1 \
@@ -21,7 +21,16 @@ OUTPUT_DIR="data/Hunyuan-Mixkit-Data"
 
 
 torchrun --nproc_per_node=$GPU_NUM \
-    ./fastvideo/data_preprocess/finetune_data_text_encoder.py \
+    fastvideo/data_preprocess/preprocess_text_embeddings.py \
     --model_type $MODEL_TYPE \
     --model_path $MODEL_PATH \
-    --output_dir=$OUTPUT_DIR 
+    --output_dir=$OUTPUT_DIR 
+
+
+
+torchrun --nproc_per_node=1 \
+    fastvideo/data_preprocess/preprocess_validation_text_embeddings.py \
+    --model_type $MODEL_TYPE \
+    --model_path $MODEL_PATH \
+    --output_dir=$OUTPUT_DIR \
+    --validation_prompt_txt $VALIDATION_PATH